ENVIRONMENTAL MICROBIOLOGY

Temporal and Spatial Distribution Patterns of PotentiallyPathogenic Vibrio spp. at Recreational Beaches of the GermanNorth Sea

Simone I. Böer & Ernst-August Heinemeyer &

Katrin Luden & René Erler & Gunnar Gerdts &

Frank Janssen & Nicole Brennholt

Received: 11 December 2012 /Accepted: 14 March 2013 /Published online: 7 April 2013# Springer Science+Business Media New York 2013

Abstract The number of reported Vibrio-related wound in-fections associated with recreational bathing in Northern Eu-rope has increased within the last decades. In order to study thehealth risk from potentially pathogenicVibrio spp. in the centralWadden Sea, the seasonal and spatial distribution of Vibriovulnificus, Vibrio parahaemolyticus, Vibrio alginolyticus andVibrio cholerae were investigated at ten recreationalbeaches in this area over a 2-year period. V. alginolyticusand V. parahaemolyticus were found to be omnipresent allyear round in the study area, while V. vulnificus occurrencewas restricted to summer months in the estuaries of the riversEms and Weser. Multiple linear regression models revealed

that water temperature is the most important determinantof Vibrio spp. occurrence in the area. Differentiated re-gression models showed a species-specific response towater temperature and revealed a particularly strong effectof even minor temperature increases on the probability ofdetecting V. vulnificus in summer. In sediments, Vibrio spp.concentrations were up to three orders of magnitude higherthan in water. Also, V. alginolyticus and V. parahaemolyticuswere found to be less susceptible towards winter temperaturesin the benthic environment than in the water, indicating animportant role of sediments for Vibrio ecology. While only avery small percentage of tested V. parahaemolyticus proved tobe potentially pathogenic, the presence of V. vulnificus duringthe summer months should be regarded with care.

Introduction

Vibrionaceae constitute an important family within theGammaproteobacteria and are common natural members ofmarine and estuarine bacterial plankton communities. At least12 species of the Vibrio genus are potentially pathogenic tohumans [1]. Particularly Vibrio parahaemolyticus and Vibriovulnificus, but also Vibrio alginolyticus and Vibrio choleraeare known as causative agents of seawater-related illnesses,such as seafood poisoning in association with gastrointestinalinfections [2–4] and infections of wounds and mucous mem-branes [5–11].

A strong link between water temperature and the numberof Vibrio spp. in seawater or shellfish and the frequency ofoccurrence of Vibrio incidences has been discovered in avast range of studies in a number of regions (e.g. [12–17]).This link was assigned both to direct temperature effects as

Electronic supplementary material The online version of this article(doi:10.1007/s00248-013-0221-4) contains supplementary material,which is available to authorized users.

S. I. Böer (*) :N. BrennholtDepartment G3—Bio-Chemistry, Ecotoxicology, Federal Instituteof Hydrology, Am Mainzer Tor 1,56068 Koblenz, Germanye-mail: Boeer@bafg.de

E.-A. Heinemeyer :K. LudenGovernmental Institute for Public Health of Lower Saxony,Lüchtenburger Weg 24,26603 Aurich, Germany

R. Erler :G. GerdtsBiological Institute Helgoland, Division of Shelf Seas SystemsEcology, Alfred Wegener Institute for Polar and Marine Research,Kurpromenade,27498 Helgoland, Germany

F. JanssenFederal Maritime and Hydrographic Agency,Bernhard-Nocht-Str. 78,20359 Hamburg, Germany

Microb Ecol (2013) 65:1052–1067DOI 10.1007/s00248-013-0221-4

well as to indirect effects by planktonic food–web interactions[18, 19]. For temperate Northern EuropeanWaters, an increas-ing number of seawater-related wound infections have beenreported since the mid-1990s, mainly during summer heatwaves. The majority of these infections occurred at the BalticSea coast such as in Denmark [20, 21], Sweden [5, 22],Finland [23] and Germany [24]; however, sporadic Vibrio-related cases were recorded upon contact with North Seawaters in the Netherlands [9, 25] and Britain [26].

An increasing number of studies accumulate evidence ofan emerging risk of Vibrio-related wound infections in highlatitudes as a consequence of climate anomalies such astemporal peaks in sea surface temperatures [27]. Due to itssemi-enclosed character, the North Sea is one of the seasmost vulnerable towards such ocean warming trends [28].Mean sea surface temperatures in the North Sea have in-creased two to four times faster than average, more than1.3 °C in the last decades [28–30]. The probability of ex-treme summers and years has more than doubled simulta-neously [29], both of which could support the spreading ofpotentially pathogenic vibrios in this region. Investigationsrecently showed that Vibrio spp. numbers, including potentialpathogens, have increased within the plankton-associated bac-terial community of the North Sea during the last half century[31].

The increasing number of bathing water-related infectionsin the Northern European seas in recent years concernedscientists, and a number of studies have been conducted inorder to gain a better understanding of Vibrio occurrence andecology in these waters (e.g. [17, 32–37]). Early work hasprovided an indication of the presence of potentially patho-genic Vibrio species at the German North Sea coast [38];however, the ecology of these organisms has not been studiedin depth for these waters. The present study is aimed atelucidating the seasonal and spatial distribution of potentiallypathogenic Vibrio spp. in the central Wadden Sea and withinthe estuaries of the rivers Ems and Weser for the firsttime. V. vulnificus, V. parahaemolyticus, V. alginolyticusand V. cholerae were quantified monthly at ten recreationalbeaches over a 2-year period in relation to environmentalconditions, using a culturing approach. Species identificationwas verified via molecular biological testing for species-specific gene targets in a number of isolates. The pathogenicityof representative V. parahaemolyticus and V. cholerae isolateswas further investigated via biochemical and molecularbiological tests. Since most documented cases in Germanywere associated with V. vulnificus, special focus was givento this organism. The main research questions were: (1) DoV. vulnificus and other potentially pathogenic Vibrio speciesoccur in this area and are there species-specific distributionpatterns? (2) What are the main environmental drivers shapingthe Vibrio community in the area? (3) Are there species-specificresponses to environmental drivers such as water temperature

and salinity? (4) Are virulent strains ofV. parahaemolyticus andV. cholerae part of the Vibrio community in this region?Sediment samples were examined in addition to watersamples in order to give a more comprehensive insight intoVibrio ecology. Although attachment to sediments and par-ticles has been shown to be one of the survival strategiesof Vibrio spp. (e.g. [15]), few environmental studies haveconsidered the sedimentary matrix (e.g. [39, 40]), thusneglecting an important component with regard to theassessment of possible health risks.

Material and Methods

Study Area and Sampling

Ten beaches (eight designated and two non-designatedbeaches) along the Central Wadden Sea coast and within theestuaries of the rivers Ems and Weser were tested for theoccurrence of potentially pathogenic Vibrio species. The sitescomprise all types of coastal waters according to the classifi-cation of the EU Water Framework Directive with the excep-tion of the euhaline Wadden Sea type (Table 1, Fig. 1).

Water and sediment samples were taken monthly betweenDecember 2009 and December 2011 by staff of the localhealth authorities. Some winter samples could not be takenbecause of ice formation; one March sample had to be re-moved from analysis because the sample containers were notadequately labelled (Table 1). Water samples were takenaccording to ISO 19458 [41]. Surface sediments were sampledaseptically either directly with the sample containers or withsterile sampling devices and decanted afterwards, dependingon submergence of the sediments. Water temperature wasmeasured in situ with portable pH meters (Dyksterhusen,Borkum: Hach Lange, type HQ11d; Norderney, Norddeich:Hach Lange, type HQ30d; Duhnen, Dorum, Wremen: WTW,type ProfiLine pH 3110; Dedesdorf, Bremerhaven, Burhave:VOLTCRAFT 300 K). Samples were transported to the labo-ratory in Aurich within 3–4 h and processed immediately uponarrival. Since cold temperatures impact V. vulnificus, whilehigh temperatures may promote growth of the organism, sam-ples were only chilled during transport on hot summer days,avoiding direct contact with cool packs.

Vibrio spp. Analyses

For the detection of Vibrio spp. in the water samples, twomethods were used. Sample volumes of 0.1 mL up to100 mL were membrane filtered (Whatman, ME 25/21STL, pore size 0.45 μm), depending on the number ofcolonies expected based on water temperature and experi-ence from preceding months. Filtration of volumes <10 mLwas augmented by the addition of a sterile NaCl solution

Temporal and Spatial Distribution of Vibrio spp. in the North Sea 1053

(Merck). The filters were transferred onto CHROMagarTM

Vibrio (CHROMagar, France) and incubated for 18–24 h at36 °C. Alternatively, sample volumes of 0.1 mL (only insummer) to 1 mL (2×0.5 mL due to the maximum capacityof the plates) were plated onto CHROMagarTM Vibrio directlywith a drigalski spatula, and incubated correspondingly.

Two approaches were also used for the detection ofVibrio spp. in sediments. Using the most probable number(MPN) technique, the sediments were diluted fourfold inbuffered peptone water (Merck, supplemented with addi-tional NaCl giving a final concentration of 1.5 % NaCl).The dilution series were incubated for 18–24 h at 36 °C,

Table 1 Overview over the sampling sites and sampling schemes

Site Coordinates Classification (coastal waters) Time ofsampling

Bathing water qualitya No of samples

°N °E Sediment Water

Dyksterhusen 53.2935 7.2291 Transitional waters.fully mixed. mesotidal

Incoming tide Good 22b, e, f 22b, e, f

Borkum 53.58765 6.65618 Euhaline exposed. fully mixed Incoming tide Excellent 24f 24f

Norddeich 53.6176 7.1493 Polyhaline. exposed. fully mixed Outgoing tide Excellent 24b 24b

Norderney 53.7017 7.1493 Polyhaline. Wadden Sea type Outgoing tide Excellent 25 25

Duhnen 53.8857 8.6352 Polyhaline. Wadden Sea type Incoming tide Excellent 25 24c

Dorum 53.7416 8.5139 Polyhaline. Wadden Sea type Incoming tide Excellent 21b, c, d, f 21b, c, d, f

Wremen 53.6460 8.4916 Transitional waters.fully mixed. mesotidal

Incoming tide Excellent 21b, c, d, f 22b, c, f

Burhave 53.58361 8.37606 Transitional waters.fully mixed. mesotidal

Incoming tide Excellent 25 25

Bremerhaven 53.3216 8.3435 Transitional waters.fully mixed. mesotidal

Incoming tide No designated beach 25 25

Dedesdorf 53.44387 8.49837 Transitional waters.fully mixed. mesotidal

Incoming tide No designated beach 25 25

a According to the requirements of the European Bathing Water Directive; results from 2011b December 2009 sample missingc January 2010 sample missingd February 2010 sample missingeMarch 2010 sample missingf December 2011 sample missing

Fig. 1 Overview over thesampling area. Mean salinitiesfor the period January 2009 toDecember 2010 at the samplingsites are depicted in greyshading (Salinity chart courtesyof Dr. Uwe Brockmann andMonika Schütt, University ofHamburg)

1054 S.I. Böer et al.

followed by sub-cultivation on CHROMagarTM Vibrio andincubation at 36 °C for additional 18–24 h. All MPN serieswere done in triplicate.

Alternatively, 60 g of the sediment sample was mixedwith 60 mL of distilled water and 60 mL of SMD (syntheticsea salt solution, Dr. Brinkmann Floramed GmbH) for30 min on a magnetic stirrer in order to extract the bacteriafrom the sediments. The sediment was left to settle,followed by removal of the supernatant. Volumes of 0.1 to10 mL of the supernatant were membrane filtered and filterstransferred onto CHROMagarTMVibrio. Filtration of volumes<10 mL was augmented by the addition of a sterile NaClsolution. In addition, 1 mL of supernatant was directly platedonto CHROMagarTM Vibrio. All plates were incubated for18–24 h at 36 °C.

Presumptive Vibrio spp. colonies were tested for oxidaseactivity using Bactident® Oxidase test strips (Merck).Randomly, colonies were microscopically examined formotility and shape. For further differentiation of presumptiveV. vulnificus and V. cholerae, pure cultures of green blue toturquoise blue colonies were grown on thiosulphate citratebile sucrose agar (TCBS, Merck; green, V. vulnificus;yellow, V. cholerae). In the case of colonies with multipleshapes and colours (e.g. lighter blue to darker blue),several colonies of each variant were sub-cultivated. Inthe case of morphological uniformity, single representativeswere picked. All colonies destined for further species identifi-cation were sub-cultivated on Columbia blood agar (Oxoid)prior to biochemical testing. All presumptive V. vulnificus andV. cholerae isolates, and randomly chosen representatives ofpresumptive V. parahaemolyticus and V. alginolyticus weresubjected to the Analytical Profile Index (API) system API20E(BioMérieux, Marcy L’Etoile, France). Verified V. choleraeisolates were further examined for O1, O139, Inaba and Ogawaserotypes via agglutination testing (antiserum ZM05 (Murex),O139 “Bengal” antiserum 294487 (Denka Seiken), polyvalentantiserum 293831 (Denka Seiken), antisera No. 3133, 2890and 3609 of the Robert Koch institute). Based on the speciesassignment, colony counts were converted to concentrations ofcolony-forming units (cfu)/100 mL water and cfu/100 g sedi-ment, respectively. According to common microbiological sur-veillance practice, the highest concentration of Vibrio spp. in asample yielded by any of the approaches was used for furtherdata analyses.

PCR Detection of Species-Specificand Virulence-Associated Gene Targets

In order to check the reliability of the species assignment viaculturing and biochemical testing, a number of Vibrio strainsthat were isolated during the study and became part of ourstrain collection were tested for species-specific and addition-ally for virulence-associated genes via PCR. V. vulnificus as

the main agent of Vibrio-related wound infections in Germanyis primarily represented in this collection. As described previ-ously [42], genomic DNA of 35 V. parahaemolyticus, 106V. vulnificus and 22 V. cholerae strains was prepared using alysozyme/SDS lysis followed by a phenol/chloroform extrac-tion and an isopropanol precipitation. All PCR reactions wereconducted in triplicates with 10 ng of template DNA for eachof the strains. The universal forward primerUtoxFwas used incombination with species-specific primers for VvtoxR, VptoxRand VctoxR, respectively [43, 44]. For V. vulnificus strains,10 ng of template DNA was used, and the PCR mixturecontained 2.5 μL Taq buffer (10×), 5 μL Taq Master PCREnhancer (5×), 10 pmol of each primer, 10 mM dNTPs and 1.5 U Taq DNA polymerase (5 Prime). UtoxR/VvtoxR frag-ments were amplified under the following PCR condi-tions: 4 min at 94 °C, 30 cycles of 94° for 30 s, 61 °Cfor 30 s, 68 °C for 30 s with a final 68 °C extension of 7 min.All V. parahaemolyticus strains were additionally screened forthe hemolysin genes tdh and trh [43, 45]. Parameters used forall V. parahaemolyticus PCRs (VptoxR/tdh/trh) were the sameas for the identification of toxR genes in V. vulnificuswith twoexceptions: annealing was performed at 62 °C for 1 min andelongation at 68 °C for 1 min. For V. cholerae, a multiplex-PCR was performed with the primer sets UtoxF/VctoxR,O139F/O139R, ctxA1/ctxA2 and O1F/O1R [43, 46, 47]. Halfa micromole of each O1 Primer and 0.125 μmol of every otherprimer were used. After a denaturation of 4 min at 94 °C,30 cycles were employed (94 °C–30 s; 59 °C–30 s, 68 °C–30 s) with an extension of 5 min at 68 °C. Resulting PCRproducts were analysed by agarose gel electrophoresis (2 %agarose; 0.5× TBE). Gels were run at 80 V for 90 min, stainedwith GelRed and visualized using the ChemiDoc XRS imag-ing System (Bio-Rad). The following reference strains wereused as positive controls: V. vulnificus DSM-10143 (VvtoxR),V. parahaemolyticus RIMD 2210633 (VptoxR and trh),V. parahaemolyticus VN-0088 (tdh), V. cholerae VN-0147(O1), V. cholerae VN-0150 (O139) and V. cholerae VN-0156(ctxA). Vibrio harveyi was used as negative control.

Kanagawa Test

The 35 V. parahaemolyticus isolates and reference strainV. parahaemolyticus DSZM 11058 were tested for theKanagawa phenomenon as described by Oberbeckmann etal. [42].

Environmental Parameters

Weather data were provided by the National MeteorologicalService (DWD). The salinity of the water samples was deter-mined according to Mohr [48] as defined in DIN 38405–1 [49]using an automatic titrator (Mettler Toledo DL55). Salinitieswere calculated according to Knudsen [50].

Temporal and Spatial Distribution of Vibrio spp. in the North Sea 1055

Sediments were characterized as follows: sediments werefreeze-dried, homogenized and particles >2 mm separatedfrom the rest of the sediment by dry sieving. The <2-mmfraction was split on a rotor sampler and one to two parts ofthe sediment grinded using a planetary mill with zirconiumvessels and beads. This sub-sample was subjected to TOCanalysis according to DIN EN 12137 [51]. Two to three partsof the remaining sediment were used to determine grain sizesvia ultrasonic sieving as described elsewhere [52]. For eachseason (21/06–20/09=summer, 21/09–20/12=fall, 21/12–20/03=winter, 21/03–20/06=spring) a representative sedimentsample was analysed for grain size distribution with the KVSsoftware (author: Dr. Johann Buss, Braunschweig/Germany,version 4.01, 20/02/1997). The sediment classification wascarried out according to Figge et al. [53]. In case of seasonalvariations in the sediment classifications, additional sedimentsamples were analysed and averages used for data processing.The percentage of clay and silt was used for statistical analyses.

Statistical Analyses

The open-source program R (R Development Core Team(2008). R: A language and environment for statistical comput-ing. R Foundation for Statistical Computing, Vienna, Austria;version 2.15.1) was used for all statistical calculations. Allbacterial parameters were log10(x+1)-transformed prior to theanalyses. In cases were bacterial levels were below the detec-tion limit of the available method, a value of zero was assignedprior to logarithmic conversion. Occasional values above de-tection limits were set to the detection limit value plus one inorder to allow differentiation for rank tests. TOC values belowdetection limit were assigned half the value of the limit. Weath-er data were selected as follows: global solar irradiance as thesum of 3 days before sampling, sunshine duration as the sum of1 week before sampling as well as cloud cover, rainfall, winddirection and wind speed as a weekly mean.

Variations in Vibrio abundance between seasons and sam-pling sites were tested for significance using the Kruskal–Wallis test. Pairwise comparisons between samples wereconducted using the kruskalmc function for Kruskal–Wallispost hoc tests on dependent variables as implemented in theR package pgirmess.

Multiple correlations between Vibrio species and allenvironmental parameters were calculated by usingSpearman’s rank correlation coefficient, and significanceswere adjusted for multiple comparisons by the Bonferronimethod [p<0.000292 (0.05/171)]. Environmental parame-ters were tested for their impact on the occurrence andabundance of individual Vibrio species via stepwise multi-ple logistic regression analyses and stepwise multiple linearregression analyses, respectively. Vibrio spp. in water andsediment were regarded separately. Regressions were con-sidered significant when the p value was <0.05. Additionally,

probabilities of the presence of individual Vibrio species as afunction of water temperature were visualized using simplelogistic regression models.

Particle Transport Model

A possible drift of V. vulnificus from the Ems estuary to theisland of Borkum in summer 2010 was checked using aparticle transport model, which is developed and applied fordrift simulations at the German Federal Maritime and Hydro-graphic Agency (BSH). The 3D, baroclinic regional oceancirculation model BSHcmod [54] calculates the three-dimensional current field as well as water level, temperature,salinity and ice cover in the North Sea and the Baltic Sea withan overall horizontal resolution of 5.5 km and 900m resolutionin the German Bight and the western Baltic Sea. The modelincludes tidal and meteorological forcing, as well as barocliniceffects due to temperature changes and varying river discharge.

Based on archived results from BSHcmod, the drift ofparticles is calculated by the particle transport modelBSHdmod.L [55]. Both model components make use ofmeteorological forcing data provided by the weather predic-tion models of the DWD.

Results

Seasonal and Spatial Distribution of Vibrio spp.

PCR testing for species-specific toxR genes verified that 35V. parahaemolyticus, 106 V. vulnificus and 21 out of 22tested V. cholerae isolates belonged to the supposed speciesassigned in API testing, thus proving the reliability of ourculturing approach. All four potentially pathogenic Vibriospecies were detected during the study period. V. alginolyticuswas by far the most frequently occurring species and could bedetected in 79 % of water samples and 94 % of sedimentsamples, respectively. The second most frequent species wasV. parahaemolyticus with 44 and 67 % of positively testedwater and sediment samples, respectively. Five percent ofwater and sediment samples contained V. vulnificus, whileV. choleraewere detected in 2 % of water samples and 4 % ofsediment samples, respectively, with all isolates belonging tothe non-O1/O139 type. Vibrio spp. were not only pres-ent more often in sediments than in water but benthicV. alginolyticus, V. parahaemolyticus, and V. vulnificuswere alsoone to three times more abundant. Mean V. alginolyticus con-centrations ranged from 1.5×103 to 2.9×105 cfu/100 g in sedi-ments and from 6×101 to 8.4×104 cfu/100 mL in water, meanV. parahaemolyticus concentrations ranged from 7.6×102

to 1.6×105 cfu/100 g in sediments and from 3.6×101 to 6.3×103 cfu/100 mL in water, and V. vulnificus concentrationsranged from 0 to 4.8×103 cfu/100 g in sediments and from

1056 S.I. Böer et al.

0 to 4.6×101 cfu/100 mL in water samples (Table S3).V. cholerae was detected only in very low numbers (0–5 cfu/100 mL; 0–7 cfu/100 g) both in sediment and water.

A strong correspondence between water temperatureand the presence and abundance of V. alginolyticus,V. parahaemolyticus and V. vulnificus was observed inthis study. Water temperature values ranged from 0 °Cin winter to 26.5 °C in summer (Table S1), with the highestwater temperatures occurring primarily in July or August.Correspondingly, the number of samples positively tested forV. alginolyticus, V. parahaemolyticus and V. vulnificus washighest in these months. Furthermore, the abundance of thesethree species significantly increased with increasing tempera-ture, while V. cholerae did not show a significant seasonalpattern (Fig. 2, for results of nonparametric post hoc tests seeTable S2). Figure 2 shows clearly that, despite the generaltrend towards elevated presence and abundance at highwater temperatures, V. alginolyticus, V. parahaemolyticusand V. vulnificus had a distinct species-specific responsetowards temperature changes. While V. alginolyticusand V. parahaemolyticus were present the whole year,V. vulnificus was only detected at temperatures between 14and 26.5 °C. Presence of V. vulnificus at water temperatures<20 °C appears to be uncommon compared to what otherstudies have shown. However, this was only true for fallsamples (Fig. 3); in most samples, a threshold of 20 °C hadto be reached before culturable V. vulnificus cells could bedetected. Once reached, V. vulnificus appeared suddenly, andcould be found for several months even at decreasing tem-peratures (Figs. 2 and 3).

V. alginolyticus and V. parahaemolyticuswere both presenteven at temperatures around freezing point. V. alginolyticus,however, was overall less sensitive towards cold temperaturesthan V. parahaemolyticus, and was more or less frequentlypresent at all sites throughout May until November (Fig. S1).Presence of V. parahaemolyticus, in contrast, followed theseasonal temperature changes with a lag of 1 to 2 months,such that the number of positively tested samples was highestin August (2010) and September (2011), respectively, whiledecreasing with decreasing water temperature. Interestingly,the impact of water temperature on both organisms was foundto be more pronounced in water than in sediment samples, allthe more for V. parahaemolyticus, indicating a protectiveeffect of the sediments at low temperatures.

V. alginolyticus, V. parahaemolyticus and V. vulnificusdid not only show species-specific responses to watertemperature, but exhibited species-specific spatial distri-bution patterns as well. Although mean salinities in thestudy area ranged from 4.1 to 27.4 psu, V. alginolyticusand V. parahaemolyticuswere ubiquitously distributed over theentire region. However, their abundance varied significantlybetween sites in both sediment and water for V. alginolyticus(K=37.4 and K=49.3, respectively; p<0.001) as well as in

sediments for V. parahaemolyticus (K=31.2; p<0.001;Fig. S2; for results of the nonparametric post hoc testssee Table S4). Highest mean V. alginolyticus and V.parahaemolyticus abundances occurred at mean salin-ities of approximately 15–17 psu, while mean cell num-bers were generally lower in brackish waters or athigher salinities. In contrast, V. vulnificus occurrencewas, with one exception, restricted to sites within theEms and Weser estuaries where the lowest mean salinities of4.1 to 17.2 psu were measured, and its abundance did not varysignificantly between these sites. Figure 4 shows the range ofsalinities (4.1–17.2 psu) and water temperatures (14–26.5 °C)at whichV. vulnificuswas present. In contrast toV. alginolyticusand V. parahaemolyticus, the sediments did not have an effecton the tolerance of V. vulnificus towards cold water tempera-tures nor did they change the acceptable salinity range. Thefindings presume, however, that high temperatures couldbroaden the salinity tolerance of the organism because positiveV. vulnificus detections at high temperatures were often relatedto high salinities, while V. vulnificus presence at lower temper-atures was connected to low salinities. However, enhancedsalinity in the Weser estuary due to reduced precipitation andriverine freshwater input in summer usually coincides withparticularly high water temperatures due to the high thermalload of the river, which could cause this trend. Figure 4 showsthat despite its preference for brackish waters, V. vulnificusmayoccasionally occur at salinities that are far beyond the salinityrange that is usually tolerated by this organism: V. vulnificusoccurred in Borkum sediment in September 2010, althoughBorkum is strongly influenced by the open North Seaand has the highest mean salinity of all sites (Table S1).Since V. vulnificus occurred at the Dyksterhusen site inJuly and August 2010 prior to its detection on Borkum1 month later, we hypothesized that V. vulnificus may havedrifted towards Borkum with freshwater currents derivingfrom the Ems estuary. This was tested with a computationalsimulation. The particle transport model, which consideredcurrent and wind regimes in this area during late summer2010, showed that V. vulnificus in Borkum may have had itsorigin in the Ems estuary (Fig. 5), suggesting that it can betransported over longer distances.

Impact of Environmental Variables on Bacterial Parameters

In a first step, interactions between environmental variablesand Vibrio spp. were analysed using the Spearman’s rankcorrelation test. Figure 6 (see also Table S5) shows a sum-mary of all statistically significant correlations. V. choleraewas the only species that was not significantly correlated toany of the parameters and is thus not represented in thefigure. The scheme clearly demonstrates that Vibrio abun-dance in sediment and water was strongly positively correlat-ed, suggesting an intense link between the Vibrio communities

Temporal and Spatial Distribution of Vibrio spp. in the North Sea 1057

in both compartments. Water temperature was the most impor-tant factor that was significantly and positively correlated to V.vulnificus, V. parahaemolyticus and V. alginolyticus abundancein sediment and water, thus mirroring the observed seasonal

distribution patterns of these three species. The weather dataexhibited various interdependences. As expected, global solarirradiance and sunshine duration were strongly correlated witheach other and showed significant negative correlations with

0

5

10

15

20

25y

WaterW

ater

tem

p. [

°C] a

0

5

10

15

20

25

y

Sediment

Wat

er t

emp

. [°C

] a

01234567

log

cfu

+1/1

00 m

L b

V.al

gin

oly

ticu

s

K=91.98,p<0.001

01234567

log

cfu

+1/1

00 g c K=91.98,p<0.001

01234567

log

cfu

+1/1

00 m

L d

V.p

arah

aem

ol.

K=83.13,p<0.001

01234567

log

cfu

+1/1

00 g e K=100.7,p<0.001

0

1

2

3

4

5

log

cfu

+1/1

00 m

L f

V.vu

lnif

icu

s K=36.0,p<0.001

0

1

2

3

4

5

log

cfu

+1/1

00 g

g K=34.27,p<0.001

0

1

2

3

4

log

cfu

+1/1

00 m

L g

V.ch

ole

rae K=12.43,p=n.s.

JanFeb

MarApr

May Jun Ju

lAug

SepOct

NovDec

0

1

2

3

4

log

cfu

+1/1

00 g g K=14.32,p=n.s.

JanFeb

MarApr

May Jun Ju

lAug

SepOct

NovDec

Fig. 2 Figure visualizing the seasonal variations in Vibrio abundances inwater (left column) and sediment (right column). The upper two plots (a)show the mean water temperatures that were measured each month whilesampling. Boxplots represent the seasonal variations in abundances of V.alginolyticus (b, c), V. parahaemolyticus (d, e), V. vulnificus (f, g), and V.cholerae (h, i), respectively. Thick bars in the boxes represent the sample

median, boxes themselves show the upper and lower quartiles, whiskersshow the range and circles represent outliers. Outliers are defined as datapoints that fall below the first quartile or exceed the third quartile by 1.5times the interquartile range. Kruskal–Wallis test statistics (K) and signif-icance levels (p) for comparison of abundances between seasons are givenin the top left corner of each sub-plot

1058 S.I. Böer et al.

cloud cover, rainfall and wind speed, while being stronglypositively correlated with water temperature. High wind speedin the North Sea is usually related to westerlies which oftenbring high amounts of rain to the region, a relationship that isreflected in the significant positive correlations between windspeed, wind direction and rainfall in Fig. 6. The positiverelationship between water temperature and westerlies in com-bination with the negative relationship between water temper-ature and wind speed appears to be contradictory; however,westerlies predominate during the warmer seasons, while east-erly winds occur mainly at winter time. The described relation-ships reflect the strong seasonal dynamics typical for the studyarea, where sunny weather and warm temperatures prevail insummer, while rainy and cloudy weather predominate in win-ter. Although individual positive correlations existed betweenhigh V. alginolyticus and V. parahaemolyticus abundance, andhigh global solar irradiance and long sunshine duration, thestrongest effect of sunshine on Vibrio spp. was mainly indirectdue to the effect of sunshine on water temperature. TOCcontents were higher in fine sediments compared to coarsesediments, as reflected in the strong correlation betweenTOC and the clay and silt content of the sediments. Figure 6shows that the abundance of benthic V. alginolyticus and V.parahaemolyticus was positively linked to these nutrient-richer sediments. Fine-grained sediments predominated inthe estuaries, while the coastal sites were mainly characterizedby a sandy sediment type, which was reflected by the signif-icant negative correlation between salinity and the clay and siltcontent of the sediments. Salinity itself was strongly depen-dent on the position of the sites and affected by dry weatherperiods as reflected in its positive link with global solarirradiance. Vibrio spp. and salinity were not significantlycorrelated.

In order to further investigate the impact of environmentalparameters on the occurrence and abundance of Vibrio spp.,stepwise multiple logistic and linear regression models weredeveloped for each Vibrio species (except V. cholerae) in waterand sediment individually. Those environmental parameters

that exhibited a strong direct or indirect interdependencewith water temperature (global solar irradiation, sunshineduration, cloud cover and wind speed) and salinity (space)in the correlation analysis were removed from further sta-tistical steps. TOC was strongly connected to the clay andsilt content of the sediments; thus, the latter was addition-ally excluded. Water temperature, salinity, TOC, wind di-rection and rainfall were kept as independent variables inthe models.

The multiple stepwise logistic regression analyses showedthat high water temperature was the crucial factor for theoccurrence of V. alginolyticus, V. parahaemolyticus and V.vulnificus in water and sediment (p<0.001, except V.alginolyticus in sediment with p<0.01; Table 2), thusunderlining the strong seasonality of Vibrio spp. occurrenceobserved in this study. Additionally, high salinity had a nega-tive influence on the presence of V. parahaemolyticus in waterand sediment (p<0.01), and a positive effect on the presence ofV. alginolyticus in sediment alone (p<0.05), which indicatescertain species-specific preferences towards salinity. Further-more, the presence of benthic V. vulnificus and V. alginolyticuswas negatively affected by westerlies (p<0.01).

The results of the predictive models greatly reflected thesepatterns. High water temperature was not only related to themere presence of V. alginolyticus, V. parahaemolyticus and V.vulnificus but had a strong positive effect on the abundance ofall three species as well, and was the dominant factor in allmodels (R2=0.03 to 0.13; Table 3). High salinity was anadditional significant factor related to low abundance of V.vulnificus in water and sediment (R2=0.01), in contrast to highconcentrations of V. alginolyticus in sediment (R2=0.03).Benthic V. alginolyticus and V. parahaemolyticus concentra-tions were significantly and positively affected by the nutrientcontents of the sediments as represented by TOC (R2=0.42and 0.45, respectively), as well as with rainfall (R2=0.01),which is likely linked to freshwater-related nutrient inputs inthe area. Rainfall also positively affected V. alginolyticusabundance in the water (R2=0.01). Furthermore, westerlies

0

2

4

6

8

10

No

of

V. v

uln

ific

us−

po

siti

ve s

amp

les

0

5

10

15

20

25

Wat

er t

emp

erat

ure

[°C

]

D J F M A M J J A S O N D J F M A M J J A S O N D

2009−2011

Fig. 3 Seasonal plot showingthe number of V. vulnificus-positive samples per month inrelation to water temperature.Bars represent the sum ofpositively tested samples (waterand sediment combined); forline and scatters, the meanwater temperature of allsampling sites was taken on amonthly basis. Arrows mark themonths in which a temperaturethreshold of 20 °C wasexceeded

Temporal and Spatial Distribution of Vibrio spp. in the North Sea 1059

had a strong negative impact on Vibrio spp. concentrations inthe sediments (R2=0.02 to 0.07) and on the abundance of V.vulnificus in water (R2=0.02), presumably due to the com-bined effect of pressing higher salinity water from the openNorth Sea into the study area, while reducing the share ofnutrient-rich riverine waters.

Water temperature was the most important explanatory var-iable in all Vibriomodels; however, the seasonal distribution of

the individual Vibrio species suggested certain species-specificresponses to changes in water temperature (Fig. 2). In order totake a closer look at these individual relationships, a simplelogistic regression model was created for each of the threeVibrio species in order to predict the probability of their detec-tion as a function of water temperature. Figure 7 is based onpresence/absence data from both the sediment and water, andclearly demonstrates that the responses of V. alginolyticus, V.parahaemolyticus and V. vulnificus to temperature increaseswere in fact individually different. The simple logistic regres-sion model of V. alginolyticus strongly reflected the goodadaptation of this organism to cold temperatures, with a 85 %probability of detecting V. alginolyticus even at water temper-atures around freezing point. Below 10 °C, even minor in-creases in water temperature strongly affected V. alginolyticuspresence in the area, and at 10 °C, the probability of detectingV.alginolyticus was already 95 %, which reflects the (almost)year-round presence ofV. alginolyticus in the study area. Above10 °C, the effect of water temperature on V. alginolyticuspresence decreases substantially. At a temperature of 25 °Chowever, 100 % of samples will probably contain culturable V.alginolyticus. In comparison, V. parahaemolyticus is quite welladapted to cold temperatures, however not as good as V.alginolyticus. At 0 °C, the probability of detecting this speciesin the study area is 60 %, thus 25 % lower than the probabilityof detecting V. alginolyticus. Below 15 °C, an increment of1 °C enhances the probability of V. parahaemolyticus de-tection by ∼1.5 %; above 15 °C this trend slows downslightly. At a temperature of 25 °C, the probability of V.parahaemolyticus presence in the study area is more than90 %, indicating that many sites, particularly those in theWeser estuary where highest summer temperatures occur,will harbour V. parahaemolyticus during summer time.

In contrast, V. vulnificus presence is strongly dependent onwarm temperatures as demonstrated in Fig. 7c, and the speciesis not likely to be present at water temperatures below 15 °C.Above 15 °C, the probability of V. vulnificus presence rapidlyincreases. In a temperature range of 15–20 °C, an increment in1 °C of water temperature increases the probability of V.vulnificus presence by ∼4 %. At water temperatures above20 °C, every 1 °C increment even leads to a tenfold increase inthe probability of V. vulnificus presence, indicating that watertemperatures >20 °C strongly promote this species with sub-stantial effects of even minor temperature changes.

Pathogenic Potential of Vibrio Isolates

V. parahaemolyticus and V. cholerae strains were furthertested for presence of virulence-associated gene targets.None of the tested V. cholerae isolates contained the O1,O139 and ctxA genes, corresponding to the results of theagglutination tests, thus indicating that all strains were non-virulent representatives. Two V. parahaemolyticus isolates

Fig. 4 Bubble plots showing the range of salinities and water temper-atures at which V. vulnificus was detected in water (a) and sediment (b)in the study area. Bubble sizes depict size classes of V. vulnificusconcentrations as described in the legends. A posteriori shaded areasvisualize the water temperature and salinity ranges that favour V.vulnificus occurrence (Borkum presented as outlier)

1060 S.I. Böer et al.

proved to have the trh gene; however, none had the tdhgene that is supposed to be associated with haemolyticcapability. In contradiction, two additional isolates that hadneither the trh nor the tdh gene showed haemolytic abilities inthe Kanagawa test, suggesting the presence of additionalhaemolysis genes.

Discussion

The results of this study showed that V. alginolyticus, V.parahaemolyticus, V. vulnificus and V. cholerae occurred inwater and sediment of recreational beaches in the centralWadden Sea and within the estuaries of the rivers Ems andWeser. The ecology of these bacteria was found to be verycomplex in this particularly dynamic ecosystem and charac-terized by distinct species-specific responses to environmen-tal determinants, such as water temperature and salinity. V.alginolyticus was by far the most prevalent species in waterand sediment, followed by V. parahaemolyticus, V. vulnificusand V. cholerae non-O1/non-O139. Earlier environmentalstudies in other parts of the North Sea region revealed com-parable Vibrio community compositions [32, 35–37, 39,56–58], suggesting that this distribution pattern could be acommon feature of Vibrio communities along the North Seacoast.

V. cholerae non-O1/non-O139 occurred only sporadicallyand exhibited neither apparent seasonal nor spatial distribu-tion patterns in this study. Furthermore, we could not iden-tify any significant environmental drivers of V. choleraeoccurrence. Since salinities were generally in a range whereV. cholerae may occur [59], viability of this species must becontrolled by some other determinants.

In contrast, V. alginolyticus, V. parahaemolyticus and V.vulnificus strongly responded to increases in water temper-ature, and followed distinct seasonal cycles in terms ofisolation frequency and abundance, corresponding to earlierobservations on Vibrio ecology [19, 60, 61]. Water tempera-ture ranged between 0 and 26.5 °C throughout the study, andwas shown to be the crucial factor governing the occurrenceand abundance of these three species in the study area, both incorrelation analyses and regression models. The frequency ofoccurrence of V. alginolyticus, V. parahaemolyticus and V.vulnificus increased with increasing temperature, and theabundance of all three species was significantly higher insummer than in winter. Nevertheless, simple logistic regres-sion models revealed a distinct species-specific response of V.alginolyticus, V. parahaemolyticus and V. vulnificus towater temperature. V. alginolyticus and V. parahaemolyticuspersisted perennially in the study area; however, V.alginolyticus was found to be much better adapted to coldwater temperatures than V. parahaemolyticus with a

Fig. 5 Graphic showing apotential V. vulnificus drift fromthe Ems estuary (positive proofin July and August 2010) toBorkum (positive proof inSeptember 2010), based onmodeling of 1,000 hypotheticalparticles. Each symbolrepresents one particle andstands for a hypothetical V.vulnificus cell. The star marksthe starting position of allparticles at the beginning ofthe model calculations,which was set to the dateof V. vulnficus detection onBorkum (September 08, 2012).Calculations were runbackwards until the date of V.vulnificus detection atDyksterhusen (August 18,2012), and the most probableposition of particles during thistime is shown

Temporal and Spatial Distribution of Vibrio spp. in the North Sea 1061

probability of the presence of the two species at freezingtemperatures of 85 and 60 %, respectively. Increases in watertemperature were accompanied by an increase in the probabilityof the presence of V. alginolyticus and V. parahaemolyticus inthe study area. However, while the effect of temperature on V.alginolyticus presence diminishes substantially at temperatures>10 °C (presumably because of other environmental factorsbecoming restrictive), V. parahaemolyticus profits considerablyfrom further temperature increases. While the highest probabil-ity of the presence of V. alginolyticus corresponds to the highestwater temperatures, maximum V. parahaemolyticus occurrencewas shown to appear with a time lag of approximately 1 to2 months, suggesting that interactions with other organismscould play an important role in V. parahaemolyticus ecologyin the study area.

In contrast, presence of culturable V. vulnificus was foundto be strongly dependent on water temperatures >14 °C.This species could be exclusively isolated at water temper-atures of 14 to 26.5 °C, supporting earlier studies thatreported isolation of V. vulnificus only at water temperaturesbetween 15 and 32 °C [4]. Interestingly, the isolation of V.vulnificus at water temperatures <20 °C succeeded only inautumn samples, while comparable spring samplesremained negative. Our results clearly show that a threshold

of 20 °C has to be reached in order to establish V. vulnificusviability in the study area. Once present, the bacterium canremain culturable for several months even at lower temper-atures without significant diminution in cell numbers, beforevanishing abruptly. V. vulnificus responds particularly tostrong to minor temperature increases when water tempera-ture is overall high. Above 20 °C, every 1 °C incrementcauses a tenfold increase in the probability of V. vulnificuspresence.

Since V. vulnificus could not be detected at water temper-atures <14 °C, the question remains as to from whichsources the species is recruited at summer time. Previousstudies suggested that V. vulnificus can withdraw into thesediments and remain in a viable but non-culturable state,when environmental conditions become unfavourable [62];however, investigations on this topic were not within thescope of this study.

In addition to the species-specific responses of the threeVibrio species to water temperature, V. alginolyticus, V.parahaemolyticus and V. vulnificus showed diverging spatialdistribution patterns that uncovered individual trends in termsof salinity preferences. A range of earlier studies in differentregions described a strong relationship between salinity and hespatial distribution ofVibrio spp. [14–16]. Although salinity in

Fig. 6 Spearman’s rank correlations between Vibrio species in sedi-ment (S) and water (W), and the environmental variables. Data of thecomplete study are covered, and variables were partly grouped forbetter overview. Only significant correlations at the Bonferroni-corrected level p≤0.000292 (0.05/171) are depicted. Thin and thickconnecting lines represent significant correlations with coefficients of0.24–0.39 and 0.4–1.0, respectively. Medium-sized lines representconnections where the correlation of at least two variables within

connected groups yielded a coefficient <0.39, but where the rest wasmore strongly connected. Lines connecting to a whole group of vari-ables show that all variables within this group were correlated with therespective parameter. Individual correlations are represented by linesdirectly connected to single variables. Connecting lines are correla-tions with at least one of these two parameters. Continuous and dashedlines represent positive and negative coefficients, respectively

1062 S.I. Böer et al.

the study was spatially and temporally variable covering abroad range from 0.6 to 33.5 psu, the overall effect of salinityon the distribution of Vibrio spp. in the study area was lessimportant than the effect of water temperature. Particularly V.alginolyticus and V. parahaemolyticus tolerated the entirerange of mean salinities between ∼4 and 27 psu and wereubiquitously distributed at all sites in the study area. Never-theless, the regression models revealed that the presence of V.alginolyticus and V. parahaemolyticus at individual timepoints was linked to species-specific salinity preferences.While high salinity values were shown to be significantlynegatively correlated to occurrence of V. parahaemolyticus,they exhibited a significant positive influence on the occur-rence and abundance of V. alginolyticus. In contrast, presenceof V. vulnificus was mainly restricted to sites within the Emsand Weser estuaries with mean salinities of ∼4 to 17 psu,

and V. vulnificus abundance in the study area was significantlyrelated to low salinity. These findings reveal a preference of V.vulnificus for brackish waters, supporting earlier descriptionsas an estuarine bacterium [11]. The present study showed,however, that V. vulnificus can occasionally occur at sites withhigh mean salinities that are usually not expected to be withinthe range tolerated by this bacterium. V. vulnificus wasdetected in sediments at Borkum beach in September 2010,a site highly influenced by the open North Sea. Results of aparticle transport model showed that wind and current regimesat that time could have caused a drift of V. vulnificus from theEms estuary to Borkum, suggesting that certain environmentalconditions can favour short-term presence of V. vulnificusoutside the estuaries.

Wind direction was determined as a significant determi-nant of Vibrio spp. in the regression models. Particularly in

Table 2 Results of the stepwise multiple logistic regression models

Estimate Std. error z value Pr(>|z|) Significance Nulldeviance

Residualdeviance

AIC

Water V. alginolyticus (Intercept) –0.86 0.30 –2.90 0.00 ** 282.32 229.76 233.76

Water temperature 0.19 0.03 6.26 0.00 ***

V. parahaemolyticus (Intercept) –1.85 0.41 –4.50 0.00 *** 315.44 262.45 268.45

Water temperature 0.17 0.03 6.29 0.00 ***

Salinity –0.05 0.02 –2.70 0.01 **

V. vulnificus (Intercept) –8.06 1.64 –4.92 0.00 *** 101.09 74.14 78.14

Water temperature 0.32 0.08 3.81 0.00 ***

Sediment V. alginolyticus (Intercept) 2.97 1.29 2.31 0.02 * 127.86 110.38 118.38

Water temperature 0.14 0.05 2.73 0.01 **

Salinity 0.07 0.03 2.22 0.03 *

Wind direction –0.14 0.07 –2.10 0.01 *

V. parahaemolyticus (Intercept) –1.85 0.41 –4.50 0.00 *** 315.44 262.45 268.45

Water temperature 0.17 0.03 6.29 0.00 ***

Salinity –0.05 0.02 –2.70 0.01 **

V. vulnificus (Intercept) –6.47 2.20 –2.93 0.00 ** 101.09 60.53 66.53

Water temperature 0.49 0.12 4.10 0.00 ***

Wind direction –0.25 0.11 –2.32 0.02 *

*p<0.05; **p<0.01; and ***p<0.001, significant

Table 3 Full predictive multiple linear regression models for log10 V. alginolyticus, log10 V. parahaemolyticus and log10 V. vulnificus in sedimentand water, respectively

Predictable variable (log10) Full multiple linear regression model R2 p

V. alg.Water =(0.10×water temp.)+(0.01×rain)+0.40 0.24 <0.001

V. alg.Sediment =(0.13×water temp.)+(0.03×salinity)–(0.06×wind direction)+(0.42×TOC)+(0.01×rain)+2.14 0.30 <0.001

V. parah.Water =(0.07×water temp.)–0.08 0.19 <0.001

V. parah.Sediment =(0.11×water temp.)+(0.45×TOC)–(0.07×wind direction)+(0.01×rain)+1.73 0.19 <0.001

V. vuln.Water =(0.03×water temp.)–(0.02×wind direction)–(0.01×salinity)+0.35 0.13 <0.001

V. vuln.Sediment = (0.04×water temp.) – (0.02×wind direction) – (0.01×salinity) + 0.30 0.12 <0.001

Significant explanatory variables determined following a stepwise forward selection procedure

Temporal and Spatial Distribution of Vibrio spp. in the North Sea 1063

the sediments, wind direction significantly influenced theoccurrence of V. vulnificus and V. alginolyticus, and theabundance of all three species, respectively. Low abundanceof V. alginolyticus, V. parahaemolyticus and V. vulnificuswas significantly connected to westerly, which usuallypresses high salinity water from the open North Sea intothe study area, while reducing the share of nutrient-richriverine waters. In contrast, rainfall had a minor positiveeffect on V. alginolyticus and V. parahaemolyticus abun-dance which can likely be linked to a higher share ofnutrient-rich freshwater in the area.

Results of this study further show that sediments play avery important role for Vibrio ecology in this temperateenvironment. Culturable estimates of V. alginolyticus, V.parahaemolyticus and V. vulnificus were generally one tothree orders of magnitude higher in sediments than in waterand were ∼105 cfu/g wet sediment in summer. Althoughprevious studies in different regions showed that sedimentscan harbour high amounts of Vibrio spp. [12, 40, 63, 64],such a high difference in the number of viable cells betweensediment and water seems to be extraordinary. The possibilityof an active benthic lifestyle of Vibrio spp. has not beenseriously discussed [12]. The sediments in general are mainlyregarded as a retreat for Vibrio under unfavourable environ-mental conditions, for example, low temperatures. Our resultsindicate that sediments in fact may exhibit a protective effecton V. alginolyticus and particularly V. parahaemolyticus atwinter time. However, with regard to the generally high num-bers of V. alginolyticus, V. parahaemolyticus and V. vulnificusin sediments reported here, it seems likely that Vibrio spp. areactive members of the benthic bacterial community. Vibriospp. can use a large variety of different carbon sources [65,66], and a significant positive relationship between TOC andthe abundance of V. alginolyticus and V. parahaemolyticus insediments observed in this study suggests that these speciescould potentially gain energy from benthic organic carbonmineralization.

In order to get a first idea of the pathogenic potential ofVibrio spp. occurring in the study area, a limited number ofV. parahaemolyticus and V. cholerae strains that were iso-lated during the study were checked for the presence ofvirulence-related genes. Thirty-five V. parahaemolyticusstrains were tested for the presence of tdh and trh, 22 V.cholerae strains for the presence of ctxA, O1 and O139,respectively. None of the V. cholerae strains was pathogenic,supporting the results of our agglutination tests. Only two V.parahaemolyticus isolates tested positive for presence of the

0 5 10 15 20 25

0.85

0.90

0.95

1.00

Water temperature [°C]

Pro

bab

ility

ofV

. alg

ino

lyti

cus

pre

sen

ce a D2=0.12

0 5 10 15 20 25

0.60

0.65

0.70

0.75

0.80

0.85

0.90

Water temperature [°C]

Pro

bab

ility

ofV

. par

ahae

mo

lyti

cus

pre

sen

ce b D2=0.04

0 5 10 15 20 25

0.0

0.2

0.4

0.6

0.8

Water temperature [°C]

Pro

bab

ility

ofV

. vu

lnif

icu

sp

rese

nce

c D2=0.32

�Fig. 7 Simple logistic regression models presenting the predicted proba-bilities for the presence of V. alginolyticus (a), V. parahaemolyticus (b) andV. vulnificus (c) in the study area as a function of water temperature.Combined data from sediment and water samples were considered forthe analyses. The test statistic D2 describes the overall model performanceand is given in the upper left corner of each sub-plot

1064 S.I. Böer et al.

trh gene, but not for tdh. Comparisons with other studiesshow that generally between 3 and 5 % of environmental V.parahaemolyticus isolates bear either of the two virulencegene markers [67, 68]. This suggests that the ratio observedhere is realistic, although more strains need to be checkedfor more reliable evidence. Presence of the trh gene aloneseems to be a common characteristic of V. parahaemolyticuscommunities in Northern European waters and has beenshown for several studies in this region [39, 69, 70]. Up tonow, only tdh-positive strains were associated with V.parahaemolyticus infections in Europe, however [68], andtdh is the gene that has been mainly attributed to haemolysis[71]. Despite this, two strains that were identified as being tdh-and trh-negative exhibited haemolytic capacity in theKanagawa test. This phenomenon has been described earlier[16], and additional genes than trh and tdh have been suggestedto be involved in V. parahaemolyticus haemolysis [67, 72]. Thesmall percentage of putatively pathogenic V. parahaemolyticusand V. cholerae isolates detected here indicates that thesespecies do not constitute a significant health risk at recreationalbeaches of the German North Sea. Interestingly, however, alltrh- and Kanagawa-positive strains were isolated from sedi-ment samples. Organisms that are able to persist in sedimentbiofilms will more likely be able to colonize other tissues suchas skin or the intestinal tract [73]; thus, the role of sediments inputting forth pathogenic clones could be an interesting aspectfor future investigations.

Although putative virulence genes of V. vulnificuswere notconsidered in this study, the mere presence of V. vulnificusshould be viewed with concern. The potential health risk bythis organism needs to be properly determined, especiallyduring the summertime. The haemolysis gene vvh has beenassociated with a large proportion of V. vulnificus isolates inNorthern European Waters [37, 39] and was even used as aspecies-specific marker [74], suggesting that environmental V.vulnificus isolates inherently carry haemolysis genes. Despitethis, V. vulnificus is generally not very virulent; however, it hasan exceptional toxicity with 30 % of wound infections endingin fatalities [4]. Common surveillance practices cannot accu-rately monitor the potential health risk emanating from V.vulnificus because the species does not correlate with faecalindicators (data not shown) and can occur at sites with excel-lent water quality [75].

In conclusion, non-virulent V. alginolyticus and V.parahaemolyticus are ubiquitously and perennially distrib-uted in water and sediment at recreational beaches of theGerman North Sea, while the putative pathogen V. vulnificusfrequently occurs in the estuaries of the rivers Ems andWeser at summertime. V. alginolyticus, V. parahaemolyticusand V. vulnificus exhibited distinct seasonal cycles, withwater temperature being the crucial factor governing thepresence and abundance of all three species. Minor addi-tional effects of salinity, wind direction and rainfall were

detected. The response of V. alginolyticus,V. parahaemolyticusand V. vulnificus to water temperature was shown to bespecies-specific with a particularly good adaptation of V.alginolyticus to temperatures around the freezing point, andan enormous increase in the probability of V. vulnificuspresence with minor temperature increments at water tem-peratures >20 °C. Vibrio spp. were not only more frequentlyisolated from sediments than from water, but also theirabundance was generally one to three orders of magnitudehigher in the benthic environment. In addition, the sedi-ments were shown to have a protective effect on V.alginolyticus and particularly V. parahaemolyticus duringthe wintertime, suggesting an important role of sedimentsfor Vibrio ecology in this dynamic temperate environment.Future studies will need to clarify the extent to which climatechange may alter the composition of the Vibrio community inthe Central Wadden Sea (particularly with regard to V.vulnificus), and whether climate change could favour theemergence and spreading of virulent representatives.

Acknowledgments This study was conducted within the Germanresearch programme KLIWAS, financed by the Federal Ministry ofTransport, Building and Urban affairs. Work at the NLGAwas furtherfinancially supported by the federal state of Lower Saxony. PCRanalyses were performed throughout the German zoonose researchnetwork VibrioNet, financed by the Federal Ministry of Educationand Research. The National Meterological Service supported ourstudy with provision of weather data. The salinity chart (Fig. 1)was provided by Dr. Uwe Brockmann and Monika Schütt (Uni-versity of Hamburg). Without the invaluable help of our col-leagues at the local health authorities of the administrativedistricts Aurich, Leer and Cuxhaven, the study would not havebeen possible. Holger Weigelt and Julia Bachtin are greatly ac-knowledged for technical assistance with Vibrio and grain sizeanalyses, respectively. Further thanks go to Enoma O. Omoregiefor proofreading the manuscript and to two anonymous reviewersfor helpful comments and suggestions for improvement.

References

1. Chakraborty S, Nair GB, Shinoda S (1997) Pathogenic vibrios inthe natural aquatic environment. Rev Environ Health 12:63–80

2. LeeWC, LeeMJ,Kom JS, Park SY (2001) Foodborne illness outbreaksin Korea and Japan studies retrospectively. J Food Prot 64:899–902

3. Su HP, Chiu SI, Tsai JL, Lee CL, Pan TM (2005) Bacterial food-borne illness outbreaks in northern Taiwan, 1995–2001. J InfectChemother 11(3):146–151

4. Oliver JD (2005) Vibrio vulnificus. In: Belkin S, Colwell RR (eds)Oceans and health: pathogens in the marine environment. Springer,New York, pp 253–276

5. Melhus A, Holmdahl T, Tjernberg I (1995) First documented caseof bacteremia with Vibrio vulnificus in Sweden. Scand J Infect Dis27(1):81–82

6. Høi L, Larsen JL, Dalsgaard I, Dalsgaard A (1998) Occurrence ofVibrio vulnificus biotypes in Danish marine environments. ApplEnviron Microbiol 64(1):7–13

7. Torres L, Escobar S, López AI, Marco ML, Pobo V (2002) Woundinfection due to Vibrio vulnificus in Spain. Eur J Clin MicrobiolInfect Dis 21:537–538

Temporal and Spatial Distribution of Vibrio spp. in the North Sea 1065

8. Schmidt U, Chmel H, Cobbs C (1979) Vibrio alginolyticus in-fections in humans. J Clin Microbiol 10(5):666–668

9. Schets FM, van den Berg HH, Demeulmeester AA, van Dijk E,Rutjes SA, van Hooijdonk HJ, de Roda Husman AM (2006) Vibrioalginolyticus infections in the Netherlands after swimming in theNorth Sea. Euro Surveill 11(45):3077

10. Sganga G, Cozza V, Spanu T, Spada PL, Fadda G (2009) Globalclimate change and wound care: case study of an off-season Vibrioalginolyticus infection in a healthy man. Ostomy Wound Mange55(4):60–62

11. Pruzzo C, Huq A, Colwell RR, Donelli G (2005) Pathogenic Vibriospecies in the marine and estuarine environment. In: Belkin S,Colwell RR (eds) Oceans and health: pathogens in the marineenvironment. Springer, New York, pp 217–252

12. Kaneko T, Colwell RR (1973) Ecology of Vibrio parahaemolyticusin Chesapeake Bay. J Bacteriol 113(1):24–32

13. Motes ML, DePaola A, Cook DW, Veazey JE, Hunsucker JC,Garthright WE, Blodgett RJ, Chirtel SJ (1998) Influence of watertemperature and salinity on Vibrio vulnificus in Northern Gulf andAtlantic Coast Oysters (Crassostrea virginica). Appl EnvironMicrobiol 64(4):1459–1465

14. Randa MA, Polz MF, Lim E (2004) Effects of temperatureand salinity on Vibrio vulnificus population dynamics asassessed by quantitative PCR. Appl Environ Microbiol70(9):5469–5476

15. Hsieh JL, Fries JS, Noble RT (2008) Dynamics and predictivemodelling of Vibrio spp. in the Neuse River Estuary, NorthCarolina, USA. Environ Microbiol 10(1):57–64

16. Martinez-Urtaza J, Lozano-Leon A, Varela-Pet J, Trinanes J, PazosY, Garcia-Martin O (2008) Environmental determinants of theoccurrence and distribution of Vibrio parahaemolyticus in theRias of Galicia, Spain. Appl Environ Microbiol 74(1):265–274

17. Oberbeckmann S, Wichels A, Wiltshire KH, Gerdts G (2011)Occurrence of Vibrio parahaemolyticus and Vibrio alginolyticus inthe German Bight over a seasonal cycle. Antonie van Leeuwenhoek100:291–307

18. Tamplin ML, Gauzens AL, Huq A, Sack DA, Colwell RR (1990)Attachment of Vibrio cholerae serogroup O1 to zooplankton andphytoplankton of Bangladesh waters. Appl Environ Microbiol56(6):1977–1980

19. Turner JW, Good B, Cole D, Lipp EK (2009) Plankton composi-tion and environmental factors contribute to Vibrio seasonality.ISME J 3:1082–1092

20. Bock T, Christensen CN, Eriksen NH, Winter S, Rygaard H,Jörgensen F (1994) The first fatal case of Vibrio vulnificus infec-tion in Denmark. APMIS 104:874–875

21. Dalsgaard A, Fridmodt-Møller N, Bruun B, Høi L, Larsen JL(1996) Clinical manifestations and molecular epidemiology ofVibrio vulnificus infections in Denmark. Eur J Clin MicrobiolInfect Dis 15(3):227–232

22. Andersson Y, Ekdahl K (2006) Wound infections due to Vibriocholerae in Sweden after swimming in the Baltic Sea, summer2006. Euro Surveill 11(8):3013

23. Lukinmaa S, Mattila K, Lehtinen V, Hakkinen M, Koskela M,Siitonen A (2006) Territorial waters of the Baltic Sea as a sourceof infections caused by Vibrio cholerae non-O1, non-O139: reportof 3 hospitalized cases. Diagn Microbiol Infect Dis 54:1–6

24. Frank C, Littman M, Alpers K, Hallauer J (2006) Vibrio vulnificuswound infections after contact with the Baltic Sea, Germany. EuroSurveill 11(8):3024

25. Mertens A, Nagler J, Hansen W, Gepts-Friedenreich E (1979)Halophilic, lactose-positive Vibrio in a case of fatal septicemia. JClin Microbiol 9:233–235

26. Reilly GD, Reilly CA, Smith EG, Baker-Austin C (2011) Vibrioalginolyticus-associated wound infection acquired in British wa-ters, Guernsey, July 2011. Euro Surveill 16(42):10

27. Baker-Austin C, Trinanes JA, Taylor NGH, Hartnell R, Siitonen A,Martinez-Urtaza J (2012) Emerging Vibrio risk at high latitudes inresponse to ocean warming. Nat Clim Chang. doi:10.1038/nclimate1628

28. Belkin IM (2009) Rapid warming of large marine ecosystems.Prog Oceanogr 81:207–213

29. MacKenzie BR, Schiedek D (2007) Daily ocean monitoring sincethe 1860s shows record warming of northern European seas.Global Change Biol 13(7):1335–1347

30. Wiltshire KH, Malzahn AM, Wirtz K, Greve W, Janisch S,Mangelsdorf P, Manly BFJ, Boersma M (2008) Resilience ofNorth Sea phytoplankton spring bloom dynamics: an analysis oflong-term data at Helgoland Roads. Limnol Oceanogr 53(4):1294–1302

31. Vezzulli L, Brettox I, Pezzati E, Reid PC, Colwell RR, Höfle MG,Pruzzo C (2012) Long-term effects of ocean warming on theprokaryotic community: evidence from the vibrios. ISME J 6:21–30

32. Bauer A, Ostensvik O, Florvag M, Ormen O, Rørvik LM (2006)Occurrence of Vibrio parahaemolyticus, V. cholerae, and V.vulnificus in Norwegian Blue Mussels (Mytilus edulis). ApplEnviron Microbiol 72(4):3058–3061

33. Eiler A, Johansson M, Bertilsson S (2006) Environmentalinfuences on Vibrio populations in northern temperate and borealcoastal waters (Baltic and Skagerrak Seas). Appl EnvironMicrobiol 72(9):6004–6011

34. Lhafi SK, Kühne M (2007) Occurrence of Vibrio spp. in bluemussels (Mytilus edulis) from the German Wadden Sea. Int JFood Microb 116(2):297–300

35. Ellingsen AB, Rørvik LM (2008) Occurrence of potentially path-ogenic Vibrio spp. in Norwegian seawater. In: Ellingsen AB (ed)Vibrio parahaemolyticus, V. cholerae and V. vulnificus in Norway,with special attention to V. parahaemolyticus. Dissertation,Norwegian School of Veterinary Science

36. Schets FM, van den Berg HHJL, Marchese A, Garbom S, deRoda Husman AM (2011) Potentially human pathogenic vib-rios in marine and fresh bathing waters related to environ-mental conditions and disease outcome. Int J Hyg EnvirHealth 214(5):399–406

37. Collin B, Rehnstam-Holm AS (2011) Occurrence and potential path-ogenesis of Vibrio cholerae, Vibrio parahaemolyticus and Vibriovulnificus on the South Coast of Sweden. FEMS Microbiol Ecol78:306–313

38. Bockemühl J, Roch K, Wohlers B, Aleksic V, Aleksic S, Wokatsch R(1986) Seasonal distribution of facultatively enteropathogenic vibrios(Vibrio cholerae, Vibrio mimicus, Vibrio parahaemolyticus) in thefreshwater of the Elbe river at Hamburg. J Appl Bacteriol 60:435–442

39. Hervio-Heath D, Colwell RR, Derrien A, Robert-Pillot A, FournierJM, Pommepuy M (2002) Occurrence of pathogenic vibrios incoastal areas of France. J Appl Microbiol 92:1123–1135

40. Vezzulli L, Pezzati E, Moreno M, Fabiano M, Pane L, Pruzzo C,The VibrioSea Consortium (2009) Benthic ecology of Vibrio spp.and pathogenic Vibrio species in a coastal Mediterranean environ-ment (La Spezia Gulf, Italy). Microb Ecol 58(4):208–818

41. Anonymous (2006) ISO 19458:2006–12. Water quality—samplingfor microbiological analysis. International Organization forStandardization, Geneva

42. Oberbeckmann S, Wichels A, Maier T, Kostrzewa M, RaffelbergM, Raffelberg S, Gerdts G (2011) A polyphasic approach for thedifferentiation of environmental Vibrio isolates from temperatewaters. FEMS Microbiol Ecol 75:145–162

43. Bauer A, Rørvik L (2007) A novel multiplex PCR for the identi-fication of Vibrio parahaemolyticus, Vibrio cholerae and Vibriovulnificus. Lett Appl Microbiol 45:371–375

44. Di Pinto A, Ciccarese G, Tantillo G, Catalano D, Forte VT (2005) Acollagenase-targeted multiplex PCR assay for identification of Vibrio

1066 S.I. Böer et al.

alginolyticus, Vibrio cholerae, and Vibrio parahaemolyticus. J FoodProtect 68(1):150–153 (4)

45. Tada J, Ohashi T, Nishimura N, Shirasaki Y, Ozaki H, Fukushima S,Takano J, Nishibuchi M, Takeda Y (1992) Detection of the thermo-stable direct hemolysin gene (tdh) and the thermostable directhemolysin-related hemolysin gene (trh) of Vibrio parahaemolyticusby polymerase chain-reaction. Mol Cell Probes 6:477–487

46. Nandi B, Nandi RK, Mukhopadhyay S, Nair GB, Shimada T,Ghose AC (2000) Rapid detection of species-specific identificationof Vibrio cholerae using primers targeted to the gene of outermembrane protein ompW. J Clin Microbiol 38:4145–4151

47. Mantri CK, Mohapatra SS, Ramamurthy T, Ghosh R, Colwell RR,Singh DV (2006) Septaplex PCR assay for rapid identification ofVibrio cholerae including detection of virulence and int SXTgenes. FEMS Microbiol Lett 265:208–214

48. Mohr CF (1856) Neue massanalytische Bestimmung des Chlors inVerbindungen. Ann Chem Pharm 97(3):335–338

49. Anonymous (1985) DIN 38405–1: 1985–12. German standardmethods for the examination of water, waste water and sludge;anions (group D); determination of chloride ions (D 1). DeutschesInstitut für Normung e.V, Berlin

50. Knudsen M (1901) Hydrographical tables. G.E.C. Gad,Copenhagen, p 63

51. Anonymous (2001) DIN EN 13137:2001–12. Characterization ofwaste—determination of total organic carbon (TOC) in waste,sludges and sediments. European Committee for Standardization,Brussels, German version

52. Ackermann F (1980) A procedure for correcting the grain sizeeffect in heavy metal analyses of estuarine and coastal sediments.Environ Technol Lett 1:518–527

53. Figge K, Köster R, Thiel H, Wieland P (1980) Schlickuntersuchungenim Wattenmeer der Deutschen Bucht. Zwischenbericht über einForschungsprojekt der KFKI. Die Küste 35:187–204

54. Dick S, Kleine E et al. (2008) A new operational circulation modelfor the North Sea and the Baltic using a novel vertical co-ordinate—setup and first results. Fifth International Conference on Euro-GOOS, Exeter, EuroGOOS

55. Dick S, Müller-Navarra S (2002) An operational oil spill model forthe North Sea and the Baltic—model features and applications. ThirdR&D Forum on High-Density Oil Spill Response, Brest, IMO

56. Van Landuyt HW, Van Hulle BMC, Fossepre JM, Verschraegen G(1985) The occurrence of Vibrio spp. at the Belgian coast. ActaClin Belg 40:12–16

57. Veenstra J, Rietra PJG, Coster JM, Slaats E, Dirks-Go S (1994)Seasonal variations in the occurrence of Vibrio vulnificus along theDutch coast. Epidemiol Infect 112:285–290

58. Ayres PA, Barrow GI (1978) The distribution of Vibrioparahaemolyticus in British coastal waters: report of a collabora-tive study. J Hyg 80:281–294

59. Singleton FL, Attwell R, Jangi S, Colwell RR (1982) Effects oftemperature and salinity on Vibrio cholerae growth. Appl EnvironMicrobiol 44(5):1047–1058

60. Kaneko T, Colwell RR (1978) The annual cycle ofV. parahaemolyticusin Chesapeake Bay. Microb Ecol 4:135–155

61. Tantillo GM, Fontanarosa M, Di Pinto A, Musti M (2004) Updatedperspectives on emerging vibrios associated with human infec-tions. Lett Appl Microbiol 39:117–126

62. Oliver JD (2010) Recent findings on the viable but nonculturablestate in pathogenic bacteria. FEMS Microbiol Rev 34(4):415–425

63. Baross J, Liston J (1970) Occurrence of V. parahaemolyticus andrelated haemolytic vibrios in marine environments of WashingtonState. Appl Microbiol 20(2):179–186

64. Blackwell KD, Oliver JD (2008) The ecology of V. vulnificus,Vibrio cholerae, and Vibrio parahaemolyticus in North Carolinaestuaries. J Microbiol 46(2):146–153

65. Thompson JR, Polz MF (2006) Dynamics of Vibrio populationsand their role in environmental nutrient cycling. In: Thompson FL,Austin B, Swings J (eds) The biology of vibrios. ASM Press,Washington, DC, pp 190–203

66. Sherr E, Sherr B (2000) Marine microbes: an overview. In:Kirchman DL (ed) Microbial ecology of the oceans. Wiley,London, pp 13–46

67. Martinez-Urtaza J, Bowers JC, Trinanes J, DePaola A (2010) Climateanomalies and the increasing risk of Vibrio parahaemolyticus andVibrio vulnificus illnesses. FoodResearch International, pp. 1780–1790

68. Baker-Austin C, Stockley L, Rangdale R, Martinez-Urtaza J(2010) Environmental occurrence and clinical impact of Vibriovulnificus and Vibrio parahaemolyticus: a European perspective.Environ Microbiol Rep 2(1):7–18

69. Robert-Pillot A, Guénolé A, Lesne J, Delesmont R, Fournier JM,Quilici ML (2004) Occurrence of the tdh and trh genes in Vibrioparahaemolyticus isolates from waters and raw shellfish collectedin two French coastal areas and from seafood imported into France.Int J Food Microbiol 91(3):319–325

70. Ellingen AB, Jørgensen WS, Monshaugen M, Rørvik LM (2008)Genetic diversity among Norwegian Vibrio parahaemolyticus. JAppl Microbiol 105(6):2195–2202

71. Nishibuchi M, Kaper JB (1995) Thermostable direct hemolysingene of Vibrio parahaemolyticus—a virulence gene aquired by amarine bacterium. Infect Immun 63:2093–2099

72. Caburlotto G, Gennari M, Ghidini V, Tafi MC, Lleo MM (2009)Presence of T3SS2 and other virulence-related genes in tdh-nega-tive Vibrio parahaemolyticus environmental strains isolated frommarine samples in the area of the Venetian Lagoon, Italy. FEMSMicrobiol Ecol 70(3):506–514

73. Pruzzo C, Vezzulli L, Colwell RR (2008) Global impact of Vibriocholerae interactions with chitin. Environ Microbiol 10(6):1400–1410

74. Han F, GeB (2010)Multiplex PCR assays for simultaneous detectionand characterization of Vibrio vulnificus strains. Lett Appl Microbiol51(2):234–240

75. Anonymous (2006) Directive 2006/7/EC of the European Parliamentand of the Council of 15 February 2006 concerning the managementof bathing water quality and repealing Directive 76/160/EEC. Off JEur Union L 64:37–51, 4.3.2006

Temporal and Spatial Distribution of Vibrio spp. in the North Sea 1067